Full text of "Typesetting African Languages"

See other formats

Typesetting
African languages

AN INVESTIGATION BY CONRAD TAYLOR

Typesetting African Languages

by Conrad Taylor

This investigation, experiments and report were originally inspired and
prompted by the author's interest in the publishing aspirations of a South
London voluntary organisation, but this research was not 'commissioned' (or
paid for) by them. I took the project on as a private one. I hope that this has
worked out to everyone's advantage, as I became fascinated by the subject -
and took my investigations further than anyone would reasonably have
anticipated (or have been prepared to paid for).

Despite that, the effort is very incomplete. It was particularly difficult to track
down clear reference sources for the character sets used for writing African
languages. Linguists seem largely uninterested in this problem. The least
ambiguous sources are language tutorials and dictionaries, but they are hard
to find. As a result, many important languages (e.g. Buganda, Kongo, Mossi,
Mandinka, Ndebele, Shona) remain undocumented here. However, the
principles remain the same.

I do not deal at all with Arabic here (it is a particularly difficult typesetting
problem, but lucrative enough to have attracted the attention of software
developers; the solutions are well documented elsewhere). I also decided not
to include the Austronesian languages of Madagascar, so the focus here is on
continental Black Africa.

Today, Black Africa is in urgent need of better means for transmitting vital
information about health issues, agricultural techniques and other means of
improving life in the African countryside and the rapidly-growing cities. I hope
that more linguists, scholars, writers, designers and software engineers will
contribute their skills to ensure that more effective means can be developed
to propagate this information effectively in the many languages of Africa.

It is my belief and hope that computers should be our salvation in this work,
and not be a part of the problem, though the issue of 'intellectual property' in
font technology does require some careful minking - and creative, generous
solutions - to ensure that Africans are not being charged more than they can
pay for the right to communicate in print in their own languages.

Contents

African languages and writing systems 1

A large quilt of small patches - 'Written African': indigenous
scripts - Alphabets introduced by missionaries.

Fonts for typesetting African languages: the issues 6

Five levels of difficulty in typesetting African languages
by computer - What is a font? And what's in it? - Standard
repertoire character sets - African typesetting using standard
fonts - 'Level 1 & 2' languages and the Internet.

Typesetting African languages with TgX 12

Introducing TpX - Glueing accents to characters - TeX in
practice (an example of Yoruba typesetting) - Preparing
the typesetting file - Typesetting the file - Some notes and
conclusions.

Obtaining modified fonts for African languages 18

Using a font editing program - Purchasing a specially-
engineered font or a 'superfont' set - the Summer Institute of
Linguistics extended-latin fonts.

The Ethiopic script system 23

Background, history and use of the Ethiopic (Ge'ez) script
system - a complete solution from EthiO Systems, illustrated
with their own Web pages.

Appendix A: Font technology notes 32

PostScript font format - Origins of TrueType - the role of
Unicode - the OpenType project.

Appendix B: Character sets for some African languages 34

Baule - Chewa, Chichewa or Nyanja - Edo or Bini - Fulfulde
or Pular - Hausa - Kikuyu - Krio - Igbo - Oromo or Galla -
Somali - Swahili - Tswana - Twi, Akan, Fante or Ashanti -
Wolof - Xhosa - Yoruba - Zulu.

African languages and writing systems

A large quilt of small patches

Africa is a continent of many indigenous languages: over 2,000 - more
than are found in any other continent. The widely- accepted classification
scheme of Joseph Greenberg divides these into four main language families
- Afro-Asiatic, Nilo-Saharan, Niger-Congo and Khoisan - the approximate
geographical distribution of which is shown on the map below.

Around 1,350 African languages are members of the Niger- Congo family,
which predominates in sub-Saharan Africa. The Bantu sub-family of 400

SENEGAL

THE GAMBIA

Fig. 1: Language families

LANGUAGE FAMILIES

Afro-Asiatic

Nilo-Saharan

Niger-Congo
Khoisan

Austronesian

Indo-European
(Altaic)

African languages and writing systems

languages was first identified as having a common origin by Wilhelm Bleek
in 1862, based on the Kongo/Luba word bantu for 'people', the equivalent
word for which in other languages of the same family is quite similar (banto,
abantu, abandu, baat, bato, vanhu etc.). Starting from a common homeland
between the Niger and Zaire river basins, the Bantu peoples have spread out
to occupy East and Southern Africa.

In contrast, a total of only 300,000 people speak languages of the small and
shrinking Khoisan family, which formerly would have been in widespread
use in Southern Africa among the hunter- gatherer peoples. Some Khoisan
languages have recently become extinct. However, the distinctive 'click'
consonants of these languages have influenced their Bantu neighbours such
as Xhosa and Zulu.

The languages of the Afro-Asiatic family predominate in North Africa. The
most prominent of these, Arabic, was imported from the Arabian peninsula
during the Muslim conquests of the seventh and eighth centuries, but there
are several highly significant indigenous Afro-Asiatic languages, especially in
the Horn of Africa and the Sahel.

The Nilo-Saharan family is probably the least 'tidy' classification, compris-
ing 200 languages. In particular there has been dispute about how to classify
Songhai, a language spoken around the Niger Bend.

(As for the Austronesian languages of Madagascar, they were brought by
colonisation from South East Asia, and are not considered in this paper.)

Only about 5% of indigenous African languages have more than a million
speakers, and only six are used by more than ten million people:

Language name

Population
(approx)

Family

Where spoken

Swahili

30 million

Niger-Congo

Tanzania, Kenya, Uganda
(lingua franca for 25m)

Hausa

25 million

Afro-Asiatic

North Nigeria,
Niger

Yoruba

20 million

Niger-Congo

Nigeria, Benin

Amharic

14 million

Afro-Asiatic

Ethiopia

(official language)

Igbo (Ibo)

13 million

Niger-Congo

Nigeria

Fula (Fulfulde)

13 million
(various dialects)

Niger-Congo

Several West African
countries

Oromo (Galla)

1 1 million

Afro-Asiatic

Ethiopia, Kenya

These six account for less than 20% of the entire population of Africa,
and the percentage would be far lower if mother-tongue speakers only were
to be counted. This is a striking contrast with South Asia (India, Pakistan,
Bangladesh, Nepal and Sri Lanka), where 17 languages have more than ten

African languages and writing systems

million mother-tongue speakers, and between them account for about 900
million people - some 70% of the combined populations of those countries.

In consequence of this linguistic fragmentation, and of long-distance trade
and colonisation, Africa has been described as a continent of lingua francas,
where Arabic and English, French and Portuguese have provided the basis
for much communication. Several African languages also function as lingua
francas, especially Swahili ('coastal language') which developed from Sabaki
dialects in East Africa but was massively influenced by Arabic and other
languages spoken by trading partners; it is now the official language of
Tanzania and is the most widely spoken single language in Uganda and
Kenya, usually as a speaker's second language.

European colonisation led to the evolution of several important Creole
languages, such as Krio in Sierra Leone. Other indigenous languages became
creolised to a degree, and were promoted as colonial powers required a
common language for their locally-recruited armies and administrations.
This was particularly so in the Belgian Congo, where Lingala became the
language of the army.

European colonial presence has also, of course, determined the setting
within which most indigenous African languages have acquired their writing
systems.

'Written African'

The traditional histories and story-telling, poetry and liturgies of almost
all African societies have been oral, not written down. 1 This may seem
ironic, when we consider that five thousand years ago the Egyptians were
among the first people to create a writing system, which was a mixture of
pictogram and alphabet.

Ancient Egypt's writing system did have some influence on later writing
systems, such as the modified hieroglyphic system used in the Kushite
empire of Meroe, but then it died out, and its inscriptions remained a
mystery until they were decoded in the early nineteenth century by Jean-
Francois Champollion.

Pure alphabets were more successful. The first fully alphabetic script was
devised around 1700 BC in northern Palestine and Syria, with 22 signs for
consonants. This gave rise to a number of different alphabetic systems,
for instance Hebrew and Arabic, Sabaean, and the script of the Phoenicians
- which was also transferred to the North African Phoenician settlement of
Carthage. The Phoenician alphabet was taken as a model by the Greeks, who
added vowels; this alphabet in turn inspired Etruscan and Roman alphabets,
and so led to the development of all the 'roman' alphabets in use today.

1 For example, a West African equivalent of the Homeric tradition is the Malian epic of the magician-king Sundiatta,
retold for centuries by Malian griots (minstrels).

African languages and writing systems

A small number of African alphabetic systems made their own separate
development from these early beginnings:

■ Tif inagh is an ancient alphabetic writing system still used today to
write Tamashek, the language of the Tuareg Berbers. It consists of
consonants only, usually written right to left, in rather square letters
made up of straight lines and dots (see Fig. 2 below). It seems to be an
ancient Libyan script derived from Carthaginian Phoenician writing
and dates from about 300 BC; rock-carved examples have been found
across North Africa and in the Canary Islands. Interestingly, Tifinagh
is used for rather domestic purposes; within the Tuareg communities,
the 'official' and written language is Arabic, in which most men but
only some women are literate.

• 0 1 n ! ■■ # i>k ■■■■ 3 $ ■■■ \\ j 1 o i ][ - x o 3 +

' bgdhwzzzntyklmnsgfqgrst
Fig. 2: Tifinagh script

■ Coptic or Old Nubian script is a modified form of the Greek alphabet,
which was used to write the Coptic language, descended from ancient
Egyptian. Coptic became extinct as a living language around 1600 AD
but continued in use in the liturgy of the Egyptian (Monophysite)
Christian church. One language which continues to use the Coptic
alphabet today is 'Nile Nubian' or Dongolawi, a Nilo-Saharan language
spoken in Egypt and Sudan by about a million people, which has a
written literature dating back to the 8th century. To support the
Nubian language, four extra consonants were added to the script.

■ Ge'ez or 'Ethiopic' script is the unique alphabet of the Horn of Africa,
developed from the old Sabaean script from the south of the Arabian
peninsula for Ge'ez, the old language of Ethiopia which survived in
liturgical use. This script was also used by the Ethiopian Jews, the
falashas, to write their scriptures. Today it is used to write three Semitic
languages of Africa: Amharic, which was promoted as the national
language of Ethiopia by Emperor Tewodros II in the 19th century;
also Tigrinya, the major language of Eritrea, and the related language
Tigre used in the north of Ethiopia. This script was originally a system
of consonants only, but has turned into a syllabary by adding an
extension to each consonant to indicate the following vowel sound.

Alphabets with a mission

Legend has it that the Greek theologian St. Cyril (827-869 ad), assisted
by his brother and fellow- missionary St. Methodius, modified the Greek
alphabet so that the Gospel could be brought to the heathen Slavs in their
own language - which had sounds for which Greek didn't have letters. Thus

African languages and writing systems

was devised the 'Cyrillic' alphabet which is used for Bulgarian, Russian and
Serbian today. Similar missionary processes developed the Roman script
so that it could be used to write Irish, Saxon and other tongues; over time,
this adaptation of the roman alphabet also led to the addition of new letters
such as y and j and w, the ligatured letterforms ft and ce and ce, and various
accents to distinguish between a much wider range of vowel sounds than
were found in Latin itself.

Essentially, that is also how most African languages have acquired their
writing systems. Just as 5th-century monks adapted the alphabet to bring
the Good News to the Angles and Saxons, latter-day missionaries devised
further modifications to the latin script to print Bibles in Yoruba and Igbo,
Gikuyu and Swahili. And this, broadly, is the origin of most of the Africa
writing systems the typesetting of which is being considered in this paper.
In fact, in studying this history I came time and time again across accounts
of how standardisation of spelling systems was slowed down by rivalry
between Catholic and Protestant promoters or alternative systems.

These details need not concern us, fascinating though they doubtless are.
However, I would like to make three points to counteract the impression
that the bringing of writing systems to Africa was entirely a missionary
endeavour:

■ In West Africa in the region of the Niger Bend and Lake Chad, societies
were involved in sophisticated trading networks across the Sahara,
centuries before Europeans anchored their ships off the Gold Coast.
For these societies, literacy was first encountered in the form of written
Arabic. Hausa is an example of a language which was written in Arabic
letters from about the 16th century, but which latterly has converted to
a latin script with some special consonants added; Swahili, used along
the East African trade routes, was also written in Arabic script in the
early 18th century.

■ In the post-colonial period, some African governments established
national commissions to reform and standardise the writing systems
and promote their use. An example of such an enterprise is the Ghana
Bureau of Languages.

■ Some of these writing systems were standardised very recently indeed.
For example, there was a great deal of controversy in Somalia about
how the language should be written, and it was a stated objective of the
1969 revolution to settle the question. One favoured contender was the
unique Osmanian alphabet, named after its inventor, Osman Yusuf.
However, the military government of Siad Barre decreed in 1972 that a
simple latin alphabet would be employed, without accents, and with
long vowels signified simply by writing the vowel twice. This decree
was followed up by an effective literacy campaign (civil servants were
given a three-month deadline to learn how to spell!) and by these
forceful means Somalia's modern writing system was established.

African typesetting issues

Fonts for typesetting African languages: the issues

Modern typesetting is done using standard personal computers,
with software of various degrees of sophistication, plus type fonts
which contain the repertoire of characters we need.

When called upon to typeset an 'unusual' language, the first issue that arises
is: do we have all the letterforms that this language requires - and does the
computer system 2 have the means to assemble them in a manner acceptable
to the users of that language? From this standpoint, I believe it is useful to
grade African languages into five grades of difficulty:

■ LEVEL 1 — these languages use only characters shared with the English
language, and also use no accents in conjunction with letters. Thus
they are extremely easy to typeset by computer.

■ LEVEL 2 — these languages do not have any specially constructed
letterforms. They do use some accents over vowels, but in a way that
is standard to common European languages such as French, Spanish or
Portuguese. This means that they can be typeset using standard fonts
and software - presenting only a slight learning difficulty, in that the
operator has to learn how to access special characters such as 6 or e.

■ LEVEL 3 — The next step up in difficulty is those languages which use
'ordinary' letterforms but in some non-standard combinations - such
as a dot under a vowel, or an acute accent over a consonant. These
languages cannot be set with standard applications and fonts. There
are two possible approaches: one is to use special typesetting software
based on 'graphic decomposition' which allows compound letterforms
to be assembled from their constituent elements; the other is to use
standard publishing software, but with specially created fonts in which
the combinations exist in ready-assembled form.

■ LEVEL 4 — These are the languages which clearly require a number
of special letterforms that do not exist in the standard fonts oriented
towards Western European language typesetting, for example the
'hooked consonants' of Hausa. Here, a special font is definitely
required, but no other modification of the system is needed.

■ LEVEL 5 — The most problematic languages have a non-latin character
set which is so large in its required repertoire that a single standard
font cannot contain them all - or perhaps they have unusual behav-
iours, such as requiring different forms of letter depending on where

2 For now, I use the general term 'computer system' so as to treat the system as a whole, without yet distinguishing the
separate contributions made by the operating system software, publishing application software, etc.

African typesetting issues

they occur in a word. This level of problem requires more than just a
special font: some other modifications will be needed, such as special
software or operating system extensions. As we are not considering
Arabic typesetting in this paper, the only script system which poses this
level of difficulty for us is the Ethiopic script system of Amharic, Tigre
and Tigrinya, for which a satisfactory solution is available if desired.

This five-level classification scheme is a useful way to assess how difficult it
would be from a technical point of view to start publishing in a particular
language. Thus, according to my investigations so far, I find that Swahili
and Somali are at Level One, Tswana is at Level Two, Igbo and Yoruba and
Nyanja are at Level Three, Twi and Krio and Hausa are at Level Four and
Amharic is at Level Five.

What is a font? And what's in it?

With some special exceptions, a font in a modern computer system is a
software resource, installed in a special relationship with the computer's
operating system 3 so that once in place, it allows the letterforms stored in
the font to be used in a wide variety of programs on the computer such
as a word-processor, DTP program or illustration program.

Internally, a modern computer font consists of a range of letterforms, each
of which is described mathematically as one or more closed paths made up
of straight lines and geometric curves; and each letterform is located within
a rectangular framework which determines the space around it. This can be
seen in the screen-shot image below.

□ ^= odieresis[246] from Minion-Yoruba H B

Fig. 3: A compound character

This is a screen capture of the editing window of
a font editing program, Fontographer, which is
often used to create new typefaces or modify
existing ones. In this case, a new combined
letterform is being created for use in typesetting
Yoruba, and is being stored in the existing slot
for the o-dieresis character (6) - which Yoruba
does not need.

Observe how the shape of the letter is defined by
mathematical curves that run between digitiza-
tion points on the letter's contour. Also note how
the letter sits in relation to its origin point and
the 'bounding box' which surrounds it.

(Fontographer does not show the letters shaded
in grey; this shading has been added afterwards
in the interests of clarity of presentation.)

3 The operating system is the most basic layer of software which a computer requires to operate, and which provided
central services to all other software. Examples of operating systems: MS-DOS, Windows, Unix, Mac OS.

African typesetting issues

Additional font data

In addition to the character data, a font will also contain tables of values
for various purposes. Hinting data provides guidance about how best to
convert outline font data to pixels for the best possible display on a screen
or printer, and kerning data provides fine adjustment to inter-letter space
for letter pairs which do not fit well together naturally.

Fonts are usually provided in matched sets known as 'families'. In a well-
constructed font family, each font is encoded with details of its 'family
membership' so that if the operator issues a simple request to switch from
the normal font to bold or italic, the correct alternate font is substituted.

The two principal formats in which type fonts can be purchased for either a
Windows or Macintosh computer are PostScript Type One and TrueType.

A brief explanation of the difference is given in Appendix A; it is not an
important distinction in terms of language support.

Standard-repertoire character sets

Within a computer system, each type character is assigned a numeric code
to identify it. In most computer systems, a single byte's worth of data is used
to store this code, and because a byte is a binary number with eight binary
'places', this gives a maximum of 256 characters to which a unique code can
in theory be assigned.

Ninety-three characters are numerically encoded identically on all systems,
a standard which was established in ASCII - the American Standard Code
for Information Interchange. This standard makes communication of
textual data possible between different programs running on the same
computer, and also between different computers, as in email applications.
This very limited characters set is satisfactory for basic English communica-
tion, and is illustrated below:

ABCDEFGHIJKLMNOPQRSTUVWXYZ

In early computing systems, the range of characters that a computer could
process was limited, because of the eight binary digits or 'bits' in each byte
of data, one was reserved for use in checking the integrity of communicated
data. However, the development of more sophisticated error- checking
schemes which did not rely on reserving a 'parity bit' in each byte means

Fig. 4: The ASCII standard
character set

This characters set is common to all
computer systems - with some old,
rare exceptions. Note that as an
American-defined standard it
includes no accented characters,
nor the pound sterling sign.

African typesetting issues

Fig. 5:

Windows font encoding

To compile this reference, an Adobe font
for Macintosh (Minion) was converted to
Windows encoding within a font editing
program. A comparison with the original
Mac encoding on the next page is most
revealing.

Note that there appear to be more 'slots'
in the Windows font than the 256 which
we would expect a byte's worth of data
to provide for, but in practice there are
only as many characters as a byte can
reference. All 32 of the initial ASCII slots
(0-31) are reserved for their original
purpose such as control codes. (All the
unoccupied slots are shaded grey here.)

Minion-Re cjulc

View by: Decimal

Name : A
Key: A

Hex: 41
Dec : 65

, ,

2 1

3 ,

[

38 || 39 || 40 || 41 || 42 || 43 || 44 || 45 || 46 || 47 || 48 || 49

(

)

1 51 II 52

■331

53 || 59 || 60 || 61

64 || 65 || 66

M.-'/M

73 || 74

79 || 80 || 81 || 82 || 83 || 84

| 85 || 86 || 87 || 88 || 89

Fil

Ell

■rl

EMI

EE!

100 || 101

[

]

| 102 || 103 || 104

105

106

,07

108

DEI

B31

■El

,24 II ,25 II ,26 ■WJBtVlBK-'ll ,30 II 131

{

■EI3

ita

■EE1

etui

141

142

143

144

■eg

■EM

■ED

■Eg

•

■TEE1

133

157

158

■ ttl

160

,62

,63

■B3

■HI

wsa

173 || 174 || 175 || 176 || 177

■HI

— i

133

ini

■i-;«i

■HI

■H4

■EEl

KiTil

Kill

wig

209

2 10

212

2,3

i'j

1441

WSfM

€

240|| 24,|| 242

243

244

245

249

250

25,

252 || 253 || 254

■35CT

■aaa

■33S1

KZ=1

that a modern computer character set can contain about 240+ characters,
which is the case in the fonts used for word processing or desktop publishing
on Windows or Macintosh computers. However, this extended character
space was implemented differently on different operating systems, as is
illustrated in Figures 5 and 6.

Fig. 5 at the top of this page shows the standard character encoding scheme
used by Adobe Systems for the fonts it supplies for use on the Windows
operating system. Note that the first 31 slots are reserved for control codes.
32 is the standard word-space, and the range from 33 to 126 constitutes the
standard ASCII character set. Character 127 is the 'delete' control code,
also reserved by ASCII. A range of extended punctuation marks, symbols,
accented vowels and other special characters required by some European
languages are deployed in most of the remaining upper slots.

African typesetting issues

Fig. 6: Mac-encoded font

In a standard Macintosh font encoding,
some built-fraction characters, and letters
required for Icelandic and East European
languages, are moved into the slots
reserved under Windows for control
characters. Without special operating
system extensions, these characters
(plus those in the range 245-255) are
rendered inaccessible to the user.

Characters in the range 128-244 are
quite easy to access on a Macintosh, due
to easy-to-remember key combinations
(such as option-e + e for e or option-a
for a).

The characters marked in colour, required
mostly for mathematical expressions, are
not actually part of each Mac font, but
are borrowed from the standard Symbol
font instead.

□

Minion-Regular i

View by: Decimal

Name R
Key: R

Hex: 41
Dec : 65

A
V

reai

i
i

■a

49 || 50

)

o
z.

54 || 55 || 56 || 57 || 58 || 59 || 60 || 6 1 || 62 || 63 || 64 || 65 || 66

UJJ

real

~C
r

| 85

88 || 89 || 90 || 91 || 92 || 93 || 94 || 95 || 96 || 97 || 98

WiTil

inn

]

WiEl

wig

■Bra

wig

wirn

,09|| ,,0|| Hill ,,2 || 113|| 114|| ,,5|| 116

WflTI

I 119 || ,20

■HI

■H4

■K-l

■Fft

,27

W3il

wai

waii

wai

Weg

waa

wen

waa

{

}

waa

WEB

WcTJ

WcEl

KTil

w?n

142 || ,43

itei

wna

KTT=1

W3il

W3a

•t

13a

Wgl

waa

■Eg

W5E1

inn

W3el

W33

■na

•

■E3

■ED

Msg

173

IfcEl

WriJ

,76

■fcU

178

179

ina

■En

182

,83

184

185

186

1 ,R7 II ■ 1 Q n|l 1 II 1Q? |l 1Q.II 1Q4 HI 10*

197

198

ra5oirwiBar2Q3i

— 1

K-I=l

K-H

PIF1

REI

215

till

F^l

R?l

F^l

FFFl

Frit

E3.

FfFl

E51

1*5=1

240

EFJ

ESI

Efl

1 255 || 256

Contrast this arrangment with Fig. 6 above, which shows the equivalent
extended encoding for the Apple Macintosh operating system, widely used
in the graphic arts. The standard ASCII characters all occupy the same slots
as in the Windows -encoded font - as of course they must. However, the
'extended' characters are deployed using different encodings.

In practice, this can and does lead to file translation errors when files are
passed from a Windows computer to a Macintosh computer or vice-versa -
for example, Windows text with typographically appropriate single quotes
'like this' would look Hike th is? when transferred to a Macintosh, and the
common Windows bullet character [•] transforms to a sigma [L] on Mac. 4

4 However, there are file translation utilities which correct for this encoding mis-match, and some desktop publishing
programs likewise re-encode the characters while importing a text file to preserve the original intended appearance.

African typesetting issues

African typesetting using standard fonts

If the reader examines the character tables in Figures 4 and 5 above in
conjunction with the tables of character use by some African languages in
Appendix B, it becomes clear that many African languages pose no special
typesetting problems because all of the characters required are provided for
in standard fonts. To refer the reader back to the five-step classification
introduced on page 6. . .

m LEVEL ONE languages which have no accents or special letters
(so can be typeset as easily as English) include Oromo, Swahili,
Somali and Zulu.

■ LEVEL TWO languages do require the use of some accented vowels,
but when the operator has figured out how to access these from within
the standard fonts there will be no problem to typeset them.

In addition, two of the important lingua francas of Africa, French and
Portuguese, are 'Level Two' languages for the purpose of this discussion.
French is widely used in e.g. Algeria, Mali, Niger, Chad, Senegal, Cameroon,
Guinea (Konakry), Cote d'lvoire, Togo, Central African Republic, Gabon,
Congo (Brazzaville), Rwanda, Burundi and the Democratic Republic of
Congo (formerly Zaire) . Portuguese is widely used in Angola, Mozambique,
Guinea-Bissau and the Cape Verde islands.

Level 1 & 2 languages and the Internet

Because all of the characters required to display Level One and Level Two
languages are in standard computer fonts, there is no difficulty using these
languages in email messages or on Web pages. The Level Two languages do
however pose something a minor problem, because of the variation between
different 'standards' for how these extended character sets are numerically
encoded. These difficulties have been resolved for the Web, and to a less
uniform degree for email users:

■ HTML encoding: to make sure that an accented character displays as
intended in all Web browsers whether on Windows, Unix, Macintosh
or other systems, it is re-encoded as a special 'character entity' within
the text of a Web page. For example, Lome would be encoded behind
the scenes as Lom&eacute ; . . . The &eacute ; fragment is displayed on
a Windows Web browser as character 233 and on a Mac as character 142
but as e in both cases.

■ Email: there are two re-encoding methods used to transfer these
characters in the body of standard email messages. The older system is
called Quoted-Printable and uses an equals sign as an escape character
followed by a two-digit hexadecimal code. Some more recent email
programs use HTML encoding.

Typesetting African languages with TEX

Using TeX to typeset African languages

An inexpensive shareware-based typesetting system popular in
academic circles can handle a broader range of African languages
than standard DTP systems. But as experiments have shown, it is
not that easy to use...

Introducing T^X

In 1977, Professor Donald Knuth of Stanford University began to investigate
the use of standard computers for typesetting complicated publications.
In particular, as a mathematician he was concerned about difficulties in
typesetting mathematical books, journals and papers, where equations are a
big problem. He devised a typesetting system called tau epsilon chi, which
are the three letters at the root of the Greek word techne (for art, or craft),
from which we get the word 'technology'. This is often typeset as Tj;X and
pronounced 'tek'.

Knuth placed TgX into the public domain, together with the METAFONT
system which he devised to make the computer typefaces which are used
by TgX typesetting systems. Hundreds of programmers, usually based at
universities, have likewise contributed their efforts to developing the TgX
typesetting system; and through this collaboration, TgX has been converted
to run on a wide range of computers - from multi-user mainframes to
personal microcomputers. The software and fonts can be dowloaded for
free, or for modest shareware fees, from a network of Internet servers
devoted to the project (the ctan archives).

Glueing accents to characters

One of the problems we have already discussed in typesetting African
languages is that diacritical marks are often required to be combined with
letters in ways that are not usual in European languages. This is a problem
for standard word-processing programs and DTP programs, because they
simply place each character to the right of the preceding one, and so they
need to have access to 'ready-composed' common combinations of letters
and diacritical marks. This means that the cedilla of c, cannot be placed
under an s or the acute accent of e be placed over an m.

TgX is different because it builds up compound accented characters
from a base character, plus floating accents. This also means that the fonts
specially designed for use with TgX are very differently organised from those
shown in figures 5 and 6 above. This can be understood better by examining
the character encoding for Computer Modern, a font designed by Knuth
himself, a PostScript equivalent of which is shown in Fig. 7 overleaf.

Typesetting African languages with TEX

Fig. 6: Computer Modern

Donald Knuth's font for use with the
TeX typesetting system is encoded very
differently from the fonts for use with
standard text composition programs.
The array of pre-composed European
accented characters found in standard
fonts is simply missing here.

Instead, TeX relies on picking up floating
accents from slots 96, 1 71 and 1 72,
246-253, 255 and 259 and composing
them together with a base character.
The full stop may also be repositioned
as an underdot character.

This approach allows a much wider
range of accented characters to be set
with TeX than with standard systems.
Note the provision of dotless i and j
(at 245 and 268) to facilitate this form
of character composition.

1 1

EDI

4£
it

%
•j

(

/
/

56 || 57 || 58 || 59 l^fTT

K
O

BE1

eei

1 i

■="■■

J. J.

Ell

-iV

EH '94

■HI

TT
U

V V

V
i

wig

BiEl

Big

Baa

Big

IHU EE;] nn

IK]

Bfl3

Bna

■DJ

■II I

CF
&

BIEI

Baa

124

125

126

127

128

129

130

131

132

134

135

136

137

138

139

140

141

142

143

144

145

1 46

150

151

152

153

154

155

157

158

159

160

161

162

163

164

165

166

■Hnl

168

169

170

■HI

■H4

■HI

176

179

180

181

184

185

186

lill

187

188

WfTil

Big

194

195

196

197

B=E1

199

200

201

202

203

204

205

muem

Kid

HI1

214

215

216

217

218

219

220

(

221

Baa

224

225

226

228

229

230

2 34

235

236

237

238

239

240

241

242

243

244

1B3

§*m

EEl

252 || 253 IBM!

PHI

H-ll

Ktfl

H*l

s|n

ffi

ffl

TeX in practice

The biggest shock about Tj;X for someone who has only recently been
introduced to text processing on a computer is that TjjX is not interactive,
and the view you get while preparing a document is definitely not 'What
You See Is What You Get' (WYSIWYG). It is a code-driven system, and
the pages are composed in a batch process.

(This in part explains the success of Tj;X within the academic community:
free from the need to support an interactive editing view of the document,
programmers can deliver parsimonious implementations of Tj;X which use
very little memory and processor power, and which can be used successfully
even from a basic terminal on a time-sharing multi-user computer.)

To explain how T£X works, we shall take the example of typesetting a short
section from modern Yoruba literature, which was done as an experiment

Typesetting African languages with TEX

Fig. 7:

A Yoruba typset sample

from A. Isola - "6 le ku"
(Ibadan, OUP, 1974)

Ord ti Ajani so wo Asake leti. 6 ni dun ri i pe ddodo ni
alaye ti Ajam' se. Inu Ajani dun. Bi 6 tile je pe ohun ti 6 le
gbe Ajani I'd hso fun Asake, sibe dna ti 6 gba gbe oro naa
kale wo ni letf pupo. Bi Asake ba le mu imdran yi 16, ati
maa lo s'ddd Ajanf kd nff sdro mo. Keke bee imu elede a
wogba. Asake ni oun a bere si i maa s'alaye org fun baba
dun, sugbdn pe diedie ni dun yid maa se e o. Ijd ti a ba gun
kd ni a rikan drun. Nwdn fi ipade si keji ni yunifasiti ni yara
Ajani.

using a Macintosh implementation of TgX (CMacTgX 3.2). The final
typeset version of the file is illustrated in Fig. 7 above.

(1 selected Yoruba for this experiment because 1 determined that, if it were
deemed acceptable to use the simple underdot character which is sometimes
used for the letters o, e and s, 5 rather than the vertical stroke which is the
alternative, then the standard Computer Modern TgX font would have all
of the components required to compose the text. 1 also chose it because
Yoruba is a significant, popular language - and a significant challenge.)

Preparing the typesetting file

A TgX typesetting file is an ordinary computer text file, using the standard
ASCII characters illustrated in Fig. 3 on page 8 above. The typesetting file
can be prepared with any simple text-editing program, on any computer,
and is afterwards processed through the TjjX software. As a simple text file,
the typesetting file can also be transferred easily to another kind of computer
- for instance by email - where it can be used to create identical output.

The TgX file contains a mixture of plain text content, and TgX 'command
words' which are preceded by a forward-slash character (\). This idea of
a mixture of text content and formatting codes will be familiar to anyone
who has worked with HTML coding to make Web pages.

For this exercise, 1 used the BBEdit 4.5 text editing program for Macintosh.
This is popular with Mac programmers and Web page creators, and is also
a good choice for Mac Tj^X-ers, because there are extensions available for
BBEdit which help by 'syntax colouring' command words so that they are
easier to distinguish during the editing process. 6

5 You may wonder how I managed to insert those characters in this text. The answer is that the FrameMaker software I
have used to prepare this report has means to move characters around from their original typeset position. However,
it is a slow an difficult process and does not offer a solution for typesetting African languages using standard fonts.

6 On a Windows system, WordPad or TextPad might be used for a similar purpose, though Word can also be used
provided that the file is saved as plain ASCII text and given a .tex file extension.

Typesetting African languages with TEX

Defining personal macro
codewords...

Unfortunately, the complexity of the accents in Yoruba meant that the
TjjX typesetting file became quite difficult to read - as can be seen below.
(Standard TgX codewords are coloured blue and comments green, author-
defined codewords are red, and line numbers have been added which were
not in the original file.)

1 %%%%%%%%%%%%%%% macros for Yoruba font characters %%%%%%%%%%%%%%

3 \def\Ed%% upper-case E with dot below

4 {E\kern- .4em\lower.45ex\hbox{ . }\kern.2em}

\def\ed%% lower-case e with dot below

{e\kern- . 35em\lower.45ex\hbox{ .}\kern . lem}

Setting margins...

Entering the text, and codes for
special characters

12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51

\def\0d%% upper-case 0 with dot below

{0\kern- . 54em\lower .45ex\hbox{ . }\kern . 24em}

\def\od%% lower-case o with dot below

{o\kern- . 38em\lower.45ex\hbox{ . }\kern . 13em}

\def\sd%% lower-case s with dot below

{s\kern- . 35em\lower.45ex\hbox{ . }\kern . lem}

\def\Aj%% character Ajani in story
{VAjVanYYU }}

\def\As%%character Asake in story
{\'A\sd Vak\'\ed{ }}

0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0^
/O/O/O/O/O/O/O/O/O/O/O/O/O/O/O/O/O/O/O/O/O/O/O/O/O/O/O/O/OA

Vaggedbottom

\baselineskip=16pt

\leftskip=100pt

\rightskip=30pt

\panndent=0pt

\hsize=4.5in

\vsize=7in

\voffset=.75in

\magnifi cation = \magstep 1

End of file.

V\Od r\'\od{} t \
lVetVYi . \'0 n\
\Aj \sd e. In\'u
ohun t\'\i{} \'o
s\'\i b\'\ed{} V
kal\'\ed{} w\od{}
V\i m\*\od r\'an
\Aj kVo n\'\i \'
imVu \ed l\'\ed
b\'\ed r\'\ed{} s
\'\od r\'\od{} fu
d\'\i \'\ed d\'\i
IjV\od{} t\'\i{}
Nw\'\od n fi ipad
ni y\~ar\'a \'Aj\
\bye

■\i{} \Aj s\od{} w\od{} \As

'\i{} Voun r\'\i{} i p\'e Vododo ni \'al\ - ay\'e t\'\i{}
\Aj dVun. B\'\i{} \'o til\'\ed{} j\'\ed{} p\'e
l\'e gbe \Aj l'\'o \'ns\od{} f\'un \'A\sd Vak\'\ed{},
\od n\'a t\'\i{} \'o gb\'a gb\'e V\od r\'\od{} n\'a\'a
ni l\'et\'\i{} p\'up\'\od{}. B\'\i{} \As b\'a le mu
y\Ai{} l\'o, Vati m\'aa l\od{} s'\'\od d\'\od{}
\i{} \sd \'oro m\"\od{}. K\'\ed k\'\ed{} b\'\ed \'\ed{}
d\'\ed{} \'a w\od gb\'a. \As n\'\i{} oun \'a
\'\i{} VYiO m\'aa \sd 'Val\'ayVe
n b\'ab\'a \'oun, \sd \~ugb\'\od n p\'e
\"\ed{} ni \~oun y\'\i \'o m\'aa \sd e \'e o.
a b\'a g\'un k\'\od{} ni a \'nkan \~\od run.
e sV\i{} kej\~\i{} n\'\i{} yunif\'as\'\i t\Ai{}
~an\'\i{}.\par

Typesetting African languages with TEX

Typesetting the file

Once the typesetting file had been prepared, the tex program was started,
and run against the prepared file. (In practice, this had to be done several
times because there were invalid commands in the file to which the tex
program objected, and these had to be tracked down and fixed.)

The result of a successful typesetting operation is a device independent
dvi file, which the tex program creates on the basis of character image data
from the METAFONT font files and the sophisticated batch-processing
algorithms for composing pages. The dvi file is a series of page images,
in which type characters are represented as black and white pixels at a
predefined resolution - which by default was 300 dots per inch. I previewed
this file without wasting paper by using the dvi preview program included
in the package.

At this stage, I could have sent the file to a laser printer, but I wanted to see if
a more versatile image of the page could be generated using the PostScript
language. The dvi2ps conversion program produced a PostScript file from
the dvi data, and because I had installed a PostScript Type One version of
the Computer Modern font in addition to the METAFONT version, the file
was created with resolution-independent characters instead of the fixed-
resolution 300 dpi ones.

I discovered that by processing the PostScript file using Acrobat Distiller

software from Adobe Systems, I could create a Portable Document Format
(pdf) file of the TjjX- typeset document. Such a file could be placed on a
Web site, to be viewed by users of Adobe's free Acrobat Reader software.
I was also able to use Acrobat software to make an Encapsulated PostScript
image file, which allowed me to place the TgX typesetting as an image on an
ordinary DTP page (which is how it made its way onto page 14).

Some notes and conclusions

■ Not every language is as troublesome to typeset in TgX as Yoruba.

In Igbo, for instance, the underdotted i, o and u characters never have
a superimposed accent, and this makes them easy to typeset using the
basic TgX codeword \d - which will place an underdot under any
desired character. But it proved impossible to precede a character with
two simple accent-placement codewords: \d Vo does not produce 6.
Thus, for Yoruba, I had to resort to a 'box placement' strategy using
the command string {o\kern-.38em\lower.45ex\hbox{.}\kern.l3em}
to generate an underdotted o. This could then be preceded by the V
grave- accent-placement command-word.

■ One can make life easier by defining one's own codewords in TjjX , and
I used this strategy here, so that the long string of codes just described
was aliased to a simple author-defined codeword, \od (see lines 12-13
in code on previous page) .

Typesetting African languages with TEX

■ Of course, a typesetting strategy which relies on this basic implement-
ation of TgX using standard METAFONT resources cannot solve the
problems of typesetting languages which have extra, special letters
that cannot be made up of existing components. Twi (Akan) and
Hausa are examples of such languages, which I therefore refer to as
level four' in my scheme of difficulties.

■ However, searches of the Cornell University Africana Web resources
indicate that some programmers have developed TgX -compatible
METAFONT fonts for African typesetting - notably Jorg Knappen
at the University of Mainz in Germany. His fc font package, which
is free shareware, is said to support Akan (Twi), Bambara, Bamileke,
Bassa, Bemba, Ciokwe, Dinka, Dholuo, Efik, Ewe-Fon, Fulani, Ga,
Gbaya, Hausa, Igbo, Kanuri, Kikuyu, Kikongo, Kpelle, Krio, Luba,
Mende, More, Nhala, Njanja, Oromo, Rundi, Kinya Rwanda, Sango,
Serer, Shona, Somali, Songhai, two systems of Sotho, Swahili, Tiv,
Yao, Yoruba, Xhosa and Zulu. 7

■ It is clear that Tj?X is a powerful and impressive typesetting system
when used to typeset materials with a very simple columnar structure
such as a textbook. Indeed, its enthusiastic supporters point out that
Tj?X has much better automated algorithms for producing pleasing
hyphenation and space distribution in the kind of formal justified-text
setting used in such publications. However, it would be very difficult,
if not impossible, to use a TgX system for the graphical and exciting
layouts required for newsletters, posters or publicity leaflets.

Therefore, while I have come away from this experiment impressed at the
capabilities of the Tj?X system, I find that I cannot recommend it at all for
the kind of typesetting and publication-design task which a voluntary
organisation would require. That needs an easy-to-use Wysiwyg system
which integrates the entry, placement and formatting of text and graphics
in an interactive editing view, and which shows the real font characters on
screen as you type - not scary formatting codes.

7 We have not been able to verify this, nor see samples of output. Of course, some of the languages in this list are
not at all problematic to typeset; others definitely are. Apparently the fc fonts were used to typeset an important
Hausa-English dictionary, for instance.

Special roman fonts for African languages

Obtaining modified fonts for African languages

An examination of the character charts in Appendix B shows that a number
of African languages require special characters and diacritical marks in
combination with characters, and standard fonts do not support these
needs. (Indeed, an illustration program had to be used to prepare those
charts, because they could not be typeset by normal means.)

Using a font editing program

One way to get the letterforms and combinations required would be to take
an existing font, open the font data package using an editing program such
as Macromedia Fontographer (as seen in figures 2, 4 and 5 above), delete
unwanted characters from the font, and replace them with the characters
required by the language in question. In many cases this editing could be
achieved by copy-and-paste methods that would pose few technical or
aesthetic challenges, for example to place a circumflex over a w character
to create the w required for Nyanja. The edited font would then be saved
as a new font 8 and could then be installed and used in the normal way.

A few points should be made about this process:

■ Levels of difficulty — Making custom assemblies of existing letter-
forms with existing accents is very easy. Creating new letterforms - as
would be required by e.g. Krio or Hausa - is more difficult, especially
as one would wish these to fit in smoothly with the rest of the letters.
The spacing arrangements between newly- created and existing letter-
forms would also need to be checked and adjusted.

■ Legality — A font is a software program, and when you 'buy' one
you in fact merely license the right to use it on a number of designated
computers and printers (see the font license from the particular font
vendor for particulars of each licene). One should be very careful to
ensure that modifying a font and re-saving it does not constitute a
breach of the licencing agreements.

In general we know that rearranging or modifying a font to which one
has a right of usage is not taken by most font vendors to be a breach
of the licence agreement, so long as the modified font is used only by
the original licensee; but to give that modified font to another party
would be a clear breach of contract. Having said that, we are not
qualified to give a detailed legal opinion on this matter and more
authoritative advice might need to be sought.

8 This could be either in PostScript Type One or TrueType format, in either Macintosh or Windows encoding, and the
Fontographer software is available for both Windows and Macintosh. See Appendix A.

Special roman fonts for African languages

Purchasing a specially-engineered font
or a 'superfont' set

One way to avoid legal problems would be to find a legitimate vendor who
could sell a valid licence to a font with the range of characters required to
typeset the language or languages in question. The problem is that support
for African languages has not been a priority for the established vendors
of quality fonts such as Adobe, Agfa, Monotype, Heidelberg etc. Either they
are unaware of the problem, or they do not want to act upon it because the
market for such fonts would be too small to justify the effort.

(Software companies also consider developing economies to be a poor risk
because unauthorised copying or 'piracy' is, understandably, most common
in these markets.)

Dalton-Maag: a font-house willing to customise

A conversation with Bruno Maag of the specialist type design company
Dalton-Maag indicates that they would be prepared to make a custom
version of any of their own existing type designs, in order to prepare it
for use in African-language typesetting.

Fig. 6

The Lexia and Pan type families from Dalton-Maag could be customised.

Lexia

ABCDEFGHIJKLMNOPQRSTU

VWXYZ7ECE&

abcdefghijklmnopqrstu

vwxyzaecefiflB.,!?

$<t£€0123456789%% 0

ABCDEFGHIJKLMNOPQRSTU

VWXYZJECE&

abcdefghijklmnopqrstu

vwxyzcecefiflfi.,!?

$<[£€0123456789%% o

ABCDEFGHIJKLMNOPQRSTU

VWXYZJECE&

abcdefghijklmnopqrstu

vwxyzaeoefiflB.,!?

$<t£€0123456789%% 0

Pan

ABCDEFGHIJKLMNOPQRSTU

VWXYZ;£CE&

abcdefghijklmnopqrstu

vwxyzaecefiflft.,!?

$t£€oi23456789%%o

ABCDEFGHIJKLMNOPQRSTU

vwxYZ^CEffifrlss.,!?
$*£€0123456789%%o

ABCDEFGHIJKLMNOPQRSTU

VWXYZECEeT

abcdefghijklmnopqrstu

vwxyzcecefiflfl!?

$H€oi2w6f89%%o

ABCDEFGHIJKLMNOPQRSTU

VWXYZy£CE&

abcdefghijklmnopqrstu

vwxyzaecefiflfs.,!?

$ < t£€oi23456789%96o

Special roman fonts for African languages

Dalton-Maag (see www.daltonmaag.com) is a small company with offices
in Brixton, most of whose work is in producing custom fonts as part of
corporate identity projects. The example best known to the public are the
fonts produced for the National Westminster Bank, as used in their leaflets
and promotional literature, and even in the interface of their cash machines.
However, Dalton-Maag have latterly put more energy into designing their
own fonts, for public licencing. Three font families are currently available,
of which Lexia and Pan are the two most suitable for general-purpose
typesetting, and therefore for modification for African languages.

A modified font for African typesetting sourced from Dalton-Maag would
be of high technical and aesthetic quality, and Bruno Maag -who is a world-
class expert in font engineering - has a number of ideas about how to make
such a font easy to use as well. However, this might be quite an expensive
option. On top of their standard font licencing fees, Dalton-Maag typically
charge £500 a day for font customisation work - though Bruno says he is
open to negotiation.

Special roman fonts for African languages

Summer Institute of Linguistics

The Summer Institute of Linguistics (see www. si 1 .org) is a US-based
organisation, which publishes an encyclopaedia of linguistics and culture
called The Ethnologue. It is my understanding that the origins of the Institute
are in Christian missionary work- which, as has been noted above, has often
been a driving force in the promotion of literacy and the development of
writing systems.

The Institute offers for sale four 'extended Latin' font families with the
purpose of assisting the typesetting of a wide range of languages. The fonts
have more characters in them than a keyboard can typically accommodate,
but a utility is provided so that one can create a custom set of the characters
to suit the job in hand, and assign easy-to-use key sequences for inputting
the characters.

The best way to illustrate the range of supported characters is to show the
sample graphics for the 'Doulos' font - somewhat like Times Roman - from
the Summer Institute of Linguistics Web site...

Fig. 7 - part A: 'Doulos' font characters from A to R

a 3 a

a a

} i

a o @

se se

AAA

@ &

■b

t i

}

B b B

B B

E L

irj

C C c

c 0

frirrj

<f 4 <L

ji 11

d D D

e 6

13 0

3 3

•?

0 0

E E E

3 3

y e

f S J

ffiffl

J 7

e (E

F g 9

9 §

p p

of Cf h

q a

H il U

H H

i i i

1 i

R' R

f i i I i 1 i i

Special roman fonts for African languages

Fig. 7 - part B: 'Doulos' S to Z, numerals and punctuation

& A J J J 13

i
j

2
J

+r» -to -ri*
Xp IS XJ

t T T T ii a ii
T I 1 1 U U

/I
4

ii fi a

If U V U U U LI

U tJ U

U Hi/ U U U U

.•4

v V A

v~ V v v Y v a

1
0

v v ■&

v A J V Y w w

i i
1

\
o

a as. uj

CO W w x x X, X

]

X x y

y y y A y y X

)

•>

Y Y Y

Y Y Y z z ? z^

■>■>

(

)

[

]

{

}

Z Z" z

3 3 3 3 3 ? z

Z Z 2

3 3

Fig. 7 - part C: 'Doulos' diacritics sample; plus pi characters

3 a a A 3 a a A a A a A.

•

a t a A a A a A a A a a

■

□

♦

❖

★

►

▼

A. a A a A. a A a a A. a A

S A a A a A a a A a A £

—>

A S A 3 a a A ^ a a A S

A a A I A. a A a A a A 1

A a a A a A a a A a A I

1—

A a A. a A. a A. 5 a a A a

A 3 a a A a a k A 3 a a

« 3 • Y 1

A a A 3 a a A. 3 a a A a

a A a A a A a A a A a A

a A a a A £ c- £ 9 a a a

□

ad^cj^daajddd
a a a a a a

□

<■

The Ethiopic script system

The only major writing system in Africa, apart from Arabic, which
does not use the Roman script at all is the ancient Ethiopic script.

A number of closely related African Semitic languages which are spoken by
a total of about 18-20 million people in Eritrea and Ethiopia - Amharic,
Tigre and Tigrinya - use variants of an ancient writing system derived from
the South Semitic Sabaean script. The advanced kingdom and culture of
Saba' (Sheba) in what is now Yemen had already had a long interplay of
influence with the Horn of Africa, and indeed legend has it that Bilqis,
Queen of Sheba ('Makeda', as she is called in Ethiopian tradition) bore to
King Solomon of the Jews a son, Menelek, who is claimed as the founder
of the royal Ethiopian dynasty.

Around the 4th century AD there was a high degree of contact, cultural
interchange and settlement between Saba' and the emerging kingdom of
Axum in the north of modern Ethiopia. Both a Sabaean script and Greek
had been in use in the area from about the fifth century BC, and in the 4th
century ad these were joined by a purely Ethiopic script similar to Sabaean,
known as Ge'ez, which came to predominate from this date onwards.

The Ge'ez language is the common ancestor of modern Tigre and Tigrinya,
and it became the language of the Ethiopian Orthodox Christian church.
Thus, although Ge'ez ceased to be spoken around the 9th or 10th centuries,
it was retained as the liturgical language of that church, and the height of
classical Ge'ez literature was between 13th and 17th centuries.

Slightly different sets of letters are used to write southerly Amharic on the
one hand, and northerly Tigre and Tigrinya on the other. Oromo, the other
major language of Ethiopia, used to be written in a form of Ge'ez script,
but is now more commonly written with a Roman alphabet.

Like Sabaean, the original Ge'ez script consisted purely of consonants, of
which it had 26. As the same script was adopted for use with the related
language of Amharic, it gained more consonants, now having a total of 33.
This process was one of gradual evolution. However, in what was probably
a conscious act of reform, the consonants were later conceived as having
seven 'orders' - depending on the vowel-sound pronounced after each - and
each letter acquired a vocalisation marker attached to it. Thus the modern
Ge'ez/ Amharic scripts have hundreds of distinct compound glyphs, and
function as a kind of 'syllabary'.

The large number of glyphs obviously can cause a problem for using a
computer to typeset the languages that use this script.

The Ethiopic script system

Fig. 8: an Amharic font:
AmharQ.ttf

This shareware TrueType font with
Windows encoding is available free from
a number of Web sites.

□

View by: Decimal

^ I Name: .notdef
' Key : **

Hex: 0
Dec 0

1 °

1 1

1:3

1 1 ?

□

Kll

46 || 47

(

)

* i'-

Ezl

■il

■±1

■*■

79 || 80

"IT

EE1

Big

BBl

■

wig

DSD.

1144

■i-m

107 j] 108 || 109 || 1 10 || 1 1 1

lit]

114 || 115 || 116 || 117

WEI

9°

WEI

B3il

FEU

B^l

B33

Bgl

127

128

129

■HI

■

B3a

141

142

143

144

B33

■E9

Baa

B31

Baa

•>

I 153 || 154 || 155 || 156~llktflbH| 159 IH&T

■m

■HI

■E3

B33

Baa

Dafi

if*

•

■ED

172 |WS1| 174 || 175

Baa

177

178

■El

Bin

Big

Bin

Bj£]

Big

-»

«V

| 187

188 || 189 || ,90

191

192

W=E1

■tpei

B=q

■EH

Rjg

Kin

Kig

wig

ED.

HIM

ESS

tWfl

WEI

6°

iaaa

KeTl

w^a

■o

I 238 || 239

EDI

Kll

KEI

250

251

252 || 253

254

<//

«J

»l

• *

1 255

256

257 || 258 |E3

26.

267

268

269

■

-V

One can get some idea of the appearance of an Ethiopic script by looking at
a Fontographer character- repertoire view of a free shareware TrueType font
for Windows called AmharQ; see above. The script has a distinct classical
form strongly influenced by the manuscript tradition, for which a broad-
edged reed pen was used with the edge held nearly horizontal. Unlike its
Sabaean ancestor, modern Ethiopic script is written from left to right.

A complete solution?

A number of free shareware Ethiopic fonts are available in TrueType format
for use with Windows. However, I believe the most complete Windows
solution is offered by the enterprising EthiO Systems Company of Houston,
Texas (www . neosof t . com). I cannot think of a better way of describing their
product offering than by reproducing some of their Web pages on the
following pages of this paper. . .

The Ethiopic script system

Software & Document

Ethiopian Computing

Ethiopicing Internet

Authorized Distributors

Home

WashRa is designed to provide services where Ethiopian script and language
users can take advantage of popular Windows 95 and Windows 3.1 software
including word processing, spreadsheets, database, presentation, multimedia,
Internet, and many other programs. Users can exchange mails, write or read
USENET news, design illustrations, and publish documents with Ethiopian text.

WashRa 3.0 introduces new features:

• A new user interface, where users can have access to all services at
once with a single mouse click. Setting application mode, configuring
the keyboard layout, or reading on-line help is easier than before.

• WashRa 3.0 comes with Enhanced KWK keyboard in additon to the
standard one. Now, users of MS Word and WordPerfect can type all
characters in Ethiopian script with out switching between the primary
and secondary fonts.

• Four fonts designed to meet your classical, modern, and Internet
based publication needs.

• Support for latest Win95 application programs including Office97,
CorelDRAW 7, Netscape Communicator 4.01, PageMaker 6.5, Adobe
Acrobat, MS Internet Explorer,...

• The online help has been revised and new features are added; more
specifically, the English section has major overhaul.

ile Edit View Insert Format Tools Tabl

•» Ethiopia Primary

ifls tip fls lfl> IBs Vp Vs

Vmkiujki '7 r, tt"T

|h.mA,Pi fllSS8 KA? 1J5 tiM-f*

Using WashRa with Word97

WashRa from EthiO Systems: page 1

The Ethiopic script system

WashRa 3.0 supports several Windows application programs.

• Word Processing:

O WordPerfect, MS Word, AmiPro, WordPro, MS Write,...

• Spreadsheets:

O Lotus 1-2-3, Excel, Quattro Pro,...

• Database:

O MS Access, Approach,...

• Presentation:

O Freelance, PowerPoint, Presentation

• Illustration:

O CorelDRAW, Adobe Illustrator, PhotoShop, Paintbrush.
PhotoWorks,...

• Publishers:

O PageMaker, Corel Ventura, QuarXpress,...

• Internet:

O Netscape Navigator, Netscape Communicator, MS Internet
Explorer, Eudora, Adobe Acrobat,...

• Multimedia:

O Director, Author ware,...

Next Section

Comments, please write to ethiosvs@neosoft.com

WashRa from EthiO Systems: page 1 part 2

The Ethiopic script system

Software & Document

Ethiopian Computing

Bhiopicing Internet

Authorized Distributors

Home

The EKWK virtual keyboard layout provides standard and application
dependent keyboard services, in which the former is used on all supported
programs and the later with MS Word and WordPerfect.

KWK Keyboard

• A basic key is used to enter character from the 1st order if it is in
the primary set of the font as shown below.

• A combination of basic and qualifier keys is used to enter
characters from 2nd to 7th order.

• Qualifier keys are

"u", "i",

'a", "y",

"e", "o"

and "/"

Qualifiers Keys

A-

(in

*"/

<•

/«

...

A glimpse of KWK keyboard table

When the user type a key from the 1st order, KWK will display the matching
character, but if the user follows the 1st order key with one of the qualifier
keys as shown above, KWK will map the two sequence of keys to the proper
character and display it.

Enhanced KWK Keyboard

The EKWK keyboard is essentially the same as KWK, but has more option to
the MS Word and WordPerfect users. Now, users can enter all characters in
the Ethiopian script with out switching fonts between primary and
secondary. This is done using the "Alt" key.

• A basic key is used to enter character from the 1st order if it is in the
primary set of the font, but "Alt + a basic key" if it is in the
secondary set.

• A combination of basic and qualifier keys is used to enter characters
from 2nd to 7th order.

• Qualifier keys are "u", "i", "a", "y", "e", "o", and "/".

WashRa from EthiO Systems: page 2 part 1

The Ethiopic script system

Basic Key

Qualifiers Keys

■

A-

rib

rh.

ill,

■1

<•

A glimpse of EKWK keyboard table

When the user type a key from the 1st order, EKWK will display the matching
character from the primary set , but if the user follows the 1st order key with
one of the qualifier keys as shown above, KWK will map the two sequence of
keys to the proper character and display it. However, if the user type "Alt +
Basic key" EKWK will display the matching character from the secondary
set .

Next Section

Comments, please write to ethiosvs@neosoft.com

WashRa from EthiO Systems: page 2 part 2

The Ethiopic script system

Software & Document

Ethiopian Computing

Ethiopicing Internet

Authorized Distributors

Home

Ethiopian Script Fonts

WashRa provides four fonts Washra , Ethiopia , Wookianos , and
YebSe . They can be used for classical and modern publishing, Web page and
artistic illustration. They are not designed based on decomposition method which
lend itself to distortion of the script. They come in TrueType form, but they are also
available in PostScript Type-1 format. A glimpse of each font:

WashRa

«T<f.-l< ViyafU flvftM/) S'l^.nf-}

•*/£v. ; i< nwi °?*ir* hi-wftfri
(who*- viynr'f ^.i.x.r 0"Ahh x y.

hl^c n+c >nn* , >'7 Morn ktc*

Wookianos

v* u M>'iii"|! mWW.'l tf-hq^ilh'* UwH'C

w n*c e&*ie : n iiari-K -writ* m i/.y" f i
ti l, Mt\nnf: u+n mivvi hvuv) h'vcA- m \uft

Ethiopia

WashRa from EthiO Systems: page 3 part 1

The Ethiopic script system

«T<J> flfm* 0BrfM»1 \ 1 i«te.t»''} h^+C
rf.y" 0°AhAi I u/.KH tflifrl h"t"t

a-*) n? 5 H"yi^r. n+c 'wms - Mono vr

>\A;>' ft/IV : «>Atf : h'i'A.

YebSe

V"Mym"fe mirUd.'l VfliLm-'t tuioti:
V/° ll+i: VXrtiW flilit V. "IKK .1 "IdW U H

Vlh K-Rh-T - UdKK ffmfi'i h«l«lh'i II? \i"M^
mil: ll+L* hlMIKI hTC* '"to* rf.Rh'f

Next Section

Comments, please write to ethiosvs@neosoft.com

WashRa from EthiO Systems: page 3 part 2

The Ethiopic script system

Software & Document

Ethiopian Computing

Ethiopicing Internet

Authorized Distributors

Home

Documentation

Reference Manual

A reference manual written in Amharic provides a detail and step by step
instructions on installation, configuration, keyboard layout, using WashRa
applications programs, Internet, Haddis Character Code, and many more.

English and Amharic On-line Help

Besides the reference manual, WashRa 3.0 comes with an on-line help
document both in English and Amharic. The on-line help includes, but not
limited to:

Introduction to WashRa 3.0,

Using WashRa with with Windows applications programs,
Entering character in Ethiopian script— KWK keyboard layout,
Tutorial on exchanging mails written based on Ethiopian script and
building database,
Troubleshooting,

Price

WashRa 3.0 $115.00 plus shipping and handling.

Owners of WashRa 2.0 can upgrade to version 3.0 for $75.00

plus shipping and handling.

Shipping and handling:

US, $4.00
Overseas, $12.00

Payment Method

Major credit cards, money order (cashier check), or wire
transfer; and check (only US).

If you prefer wire transfer, please contact us for more
information.

Comments, please write to ethiosvs@neosoft.com

WashRa from EthiO Systems: documentation & ordering

Appendix A: type technology notes

Appendix A: Type technology notes

PostScript font format

The current phase in computerised publishing was initiated in California
in the mid 1980s through close co-operation between Apple Computer and
Adobe Systems. Apple's Macintosh was the first affordable computer which
could display pages on screen as they would appear on the printer, but
the first fonts for the Mac were crude bit-maps, appropriate only for their
ImageWriter dot-matrix printer.

Meanwhile, Adobe had created the PostScript page description language,
by means of which an image of a page could be sent to a printing device,
irrespective of whether that device was a medium-resolution office laser
printer or a high-resolution graphic arts imagesetter; they also devised the
Type One scaleable outline fonts to work inside a PostScript workflow.

In 1985, Apple licensed the PostScript technology for its first laser printer,
together with a number of Adobe's PostScript Type One fonts. Linotype
also licensed the system for its laser imagesetters. Provided with this tech-
nical base, Macintosh dtp programs such as PageMaker and QuarkXPress
quickly swept away previous methods of typesetting machines.

Origins of TrueType

However, a strong faction within Apple felt that the company was unduly
reliant on Adobe's technology - for which Adobe charged hefty licensing
fees - and quietly planned an alternative outline font format based on a
different kind of geometric curves. 9 This became the TrueType font format.
Apple entered into an agreement with Microsoft whereby both companies
would support the rendering of TrueType fonts by their operating systems,
and Microsoft would provide a page description language - a PostScript
alternative - to work with the new font format.

When Apple and Microsoft went public about their TrueType initiative,
some observers thought that Adobe's Type One format was doomed,
especially as the Apple and Microsoft operating systems were upgraded
to output high-quality font data both to the screen and to low-cost printers
such as ink-jets. However, within the graphic arts industry, publishers had
come to rely on PostScript workflows to publish their magazines and adverts
and make money - and it was difficult to use TrueType fonts within such
workflows. Adobe also improved their position by creating Adobe Type
Manager (ATM), a system extension for Windows and Macintosh which
renders PostScript Type One font data nicely to screen and to low-cost

9 PostScript Type One fonts have their outlines defined in cubic-equation curves; TrueType outlines are quadratic curves.

Appendix A: type technology notes

printers, and which is bundled with desktop publishing and other graphic
arts programs. Therefore, in practice, both TrueType and PostScript Type
One font formats have survived; the former are used mostly in business
communication and the home, and the latter in professional publishing.

Does it matter whether one uses TrueType or PostScript fonts for publishing
projects? Recent technical developments in PostScript interpreter software
for imagesetters mean that it should not - but former bad experiences with
trying to use TrueType fonts in a graphic arts workflow mean that most
graphics professionals are still unwilling to use TrueType fonts in projects
which will be sent out for professional imagesetting and printing. They may
be wrong, and probably now are, but printers are known to be difficult to
separate from their technical prejudices.

Beyond eight bits: the role of Unicode

The Unicode Consortium is a project which aims to give each character in
each of the world's languages a unique reference code of its own, as a better
way of allowing computer systems to reference large character sets. The idea
is to use not just one byte's worth of data to reference a character, but two;
with sixteen binary digits, this creates a 'reference space' to refer to more
than 64,000 characters.

Windows NT4.0 was the first widely available computer operating system
to use Unicode encoding for referencing characters, and Unicode is also
supported by the Windows 2000 operating system and the Microsoft
Office 2000 application suite. Adobe InDesign, Adobe Illustrator and
QuarkXPress 5.0 are three document preparation systems likely to be
early supporters of Unicode.

Unfortunately, not all of the characters required by African languages
appear to have been yet indexed by the Unicode consortium and it is not
clear what impact Unicode will have on the develop of African typesetting.

The OpenType project

Collaboration between Adobe Systems and Microsoft Typography has
gone into developing a next-generation' font format which would be able
to contain a larger number of characters. The OpenType format may use
either PostScript's Bezier curves or TrueType's quadratics to describe the
character outlines, and will have extensive sets of tables to control the
relationships between letters.

Once publishing applications are developed which support the feature,
OpenType's glyph substitution features may be of great interest for type-
setting African languages because this will allow a sequence of key-presses
to be 'collapsed' or 'fused' into the presentation of a single composite glyph.
Thus one might press the key sequence \ + " + o and be presented with 6
on screen.

Appendix B: Character sets for African languages

The charts following this page are intended to show the
Roman letterforms and accents required to typeset a variety
of African languages. (However, it should be noted that it was
hard to find trustworthy sources of information, and more
diligent research is required to improve upon these findings.)

The language charts are presented in alphabetical order,
without page numbers.

■ Baule

■ Chewa, Chichewa or Nyanja

■ Edo or Bini

■ Fulfulde or Pular

■ Hausa

■ Kikuyu

■ Krio

■ Igbo

■ Oromo or Galla

■ Somali

■ Swahili

■ Tswana

■ Twi, Akan, Fante or Ashanti

■ Wolof

■ Xhosa

■ Yoruba

■ Zulu

Baule

Baule is a member of the Kwa sub-group of the Niger-Congo family of languages. It is
spoken by some 1-5 million people in Cote d'lvoire, and half a million people in Ghana.

Consonants

Bb Dd Ff Gg Kk
LI Mm Nn Pp Ss
Tt Vv WwYy Zz

Vowels

Aa Ee £e Ii Oo
Oo Uu

# Baule is not a difficult language to typeset by computer, provided one has access to a special font with
the correct letterforms. It should be possible to map these consonants to existing keys on the keyboard
which are not required for other purposes.

Chichewa or Nyanja

The language variously called Chewa, Chichewa or Nyanja is a member of the Bantu sub-
group of Benue-Congo languages, with a tradition of origin in the Zaire basin. It is spoken
by 3.8 million people in south-east Africa, notably in Malawi, and in the south east of
Zambia, where it is the second most common language after Bemba.

Consonants

Bb Cc Dd Ff Gg Hh
Jj Kk LI Mm Nn Pp
Qq Rr Ss Tt Vv Ww
Ww Xx Yy Zz

Vowels

Aa Ee Ii Oo Uu

Further notes

• I have marked the consonants Q, V and X in lighter grey above, as I am not certain that they are
used in Nyanja.

# The only letterform required by Nyanja that is not in the standard Macintosh or Windows fonts is
W with superscript circumflex. It may be of interest to note that this is a letterform also required
by Welsh - for which it is possible to obtain some modified fonts.

0 There are two solutions for typesetting Nyanja. If a modified font (as for Welsh) is available, any

DTP or word-processing program can be used to typeset it. Alternatively, one could use a typesetting
system which can place a floating accent on top of any arbitrarily chosen character (the 'composed
character' approach). Any implementation of TjX could do this with ease, but complex document
layouts such as leaflets and newsletters are hard to do in T[X.

Edo or Bini

Edo (or Bini) is a member of the Niger-Congo family of languages, spoken in Nigeria on
the West bank of the Niger south of the confluence with the River Benue, and in Benin.
Estimates of the number of speakers vary widely, up to 2-5 million.

Consonant range and sequence

b d f g gb gh h k kh kp 1 m mw n p
r rh rr s t v vb w y z

Consonantal letterforms

Bb Dd Ff Gg Hh Kk
LI Mm Nn Pp Rr Ss
Tt Vv Ww Yy Zz

Vowels

AatAaAa] EetEeEe] EetEeEe]

• • ....

Ii [Ii ff] 00 [06 06] 00 [06 06]

• • ....

Uu [Uu Uu]

Further notes

# In addition to the seven vowels shown above, Edo also has five nasalised wowels, but these are
signalled simply by adding an n - e.g. an.

# Edo has tonal accents (signalled with acute and grave diacriticals), but the practice is to use these only
in the hundred or so words where the lack of an accent would result in ambiguity.

# There are two possible approaches to typesetting Edo. Any implementation of TrX could do the job,
but complex document layouts such as leaflets and newsletters are hard to do in TjX. For use with
standard DTP or word-processing programs, a font with an extended character set would be preferred.

Fulfulde or Pular

Fulfulde (Pular, Pullaar, Pulle) is the language of the Fulani or Fulbe people, who are widely
dispersed throughout West Africa in a zone from Senegal to Cameroon. Related to Wolof
and Serer, Fulfulde is a member of the West Atlantic sub-group of the Niger-Congo family
and may have as many as 15 million speakers. It is a national language in Guinea, Mali and
Niger. Early adopters of a pastoral lifestyle, the Fulani have also played an important role in
the dispersion of Islam in West Africa.

Consonants

Bb *B

DcC

Vowels

Further notes

# V is not required for Fulfulde, but modified B, D, N and Y letterforms are required for extra
consonants.

# Hausa is not a difficult language to typeset by computer, provided one has access to a special font with
the correct letterforms. It should be possible to map these consonants to existing keys on the keyboard
which are not required for other purposes.

Hausa

Hausa is by far the most widely spoken member of the Chadic sub-group of Afro-Asiatic
languages, and the only one to have a written literature. Arabic script ('ajami') was intro-
duced in the 16th century, but now a modified Roman alphabet is used. About 25 million
people speak Hausa as their mother tongue, in south Niger and northern Nigeria, and
several million more speak Hausa as a second language. In Nigeria, the Hausa-speaking
Muslim community is politically influential.

Consonants

Bb *B Cc Dd Dcf Ff
Gg Hh Jj Kk Kk LI
Mm Nn Rr Ss Tt Ww
Yy Zz

Vowels

Aa Ee Ii Oo Uu

Further notes

0 Q, V and X are not required for Hausa, but modified B, D and K letterforms are required for three
glottalised consonants. (A fourth, TS, can be written with existing letterforms.) P is occasionally met
as a non-standard representation of F - which in Hausa has a pronunciation closer to P.

# Hausa has both long and short vowels. As an aid to pronunciation in learning Hausa, a macron is
sometimes used over a vowel (6) to show when it is long. However, long and short vowels are not
distinguished thus in everyday written Hausa. Similarly, Hausa is a partially tonal language, with
three tones: low, falling and high. A low tone may be indicated by a grave mark over a vowel (6)
and a falling tone with a circumflex (6), but these are not used in everyday written Hausa.

# In summary, Hausa is not a difficult language to typeset by computer, provided one has access to a
special font with the correct letterforms. It should be possible to map these consonants to existing
keys on the keyboard which are not required for other purposes.

Igbo

Igbo (or Ibo) is a member of the Niger-Congo family of languages, variously classified with
the Bantu or Kwa language sub-groups. It is one of the chief literary and cultural languages
of southern Nigeria, and is spoken by about 12 million people. In the past there have been
rival writing systems for Igbo sponsored by Catholic and Protestant missionaries; the system
now used was set out in 1961 by S. E. Onwu.

Consonants

Bb Cc Dd Ff Gg Hh
Jj Kk LI Mm Nn Pp
Qq Rr Ss Tt Vv Ww
Xx Yy Zz

Vowels

Aa Ee Ii Ii Oo Oo

• • • •

Uu Uu

Further notes

# At the time of writing it is not clear whether all of the standard Latin consonants shown are actually
required for Igbo, as the information was obtained from a reference source listing only the language's
extended Latin font requirements.

• There are two possible approaches to typesetting Igbo. Any implementation of Tr^< could typeset Igbo
with ease, using the \d control-word to position the underdots; however, complex document layouts
such as leaflets and newsletters are hard to do in TrX For use with standard DTP or word-processing
programs, a font with an extended character set would be preferred.

Kikuyu

Kikuyu is an easterly member of the Bantu sub-family of Niger-Congo languages, spoken in
Kenya by about five million people between Nairobi and Mt. Kenya. The Kikuyu were very
active in the Kenyan independence struggle and the language is politically influential.

Consonants

Bb Cc Dd Ff Gg Hh
Jj Kk LI Mm Nn Pp
Qq Rr Ss Tt Vv Ww

Vowels

Aa Ee Ii Ii Oo Uu Uu

Further notes

# There is no X or Z in Kikuyu. The letters F, L, P and V are also unused in Kikuyu, except when spelling
words of foreign origin that require them.

• There are two possible approaches to typesetting Kikuyu. Any implementation of TfX. could do the job,
but complex document layouts such as leaflets and newsletters are hard to do in T[X. For use with
standard DTP or word-processing programs, a font with an extended character set would be preferred.

Kikuyu

Consonants

Bb Cc Dd Ff Gg Hh
Jj Kk LI Mm Nn Pp
Qq Rr Ss Tt Vv Ww

Vowels

Aa Ee Ii Ii Oo Uu Uu

Further notes

# There is no X or Z in Kikuyu. The letters F, L, P and V are also unused in Kikuyu, except when spelling
words of foreign origin that require them.

Krio

Krio is an English-facing Creole language, spoken and written by approximately 350,000
people in Sierra Leone. Most of the vocabulary is recognisably derived from English.

Consonants

Bb Cc Dd Ff Gg Hh
Jj Kk LI Mm Nn Pp
Qq Rr Ss Tt Vv Ww
Xx Yy Zz

Vowels

Aa Ee £e Ii Oo
Oo Uu

Further notes

# Three tones can be distinguished in Krio - low, high and falling - and these are sometimes marked in
reference books with grave (6), acute (e) and circumflex (6) accents over the vowels. But these accents
are not employed in everyday usage.

# Krio is not a difficult language to typeset by computer, provided one has access to a special font with
the correct letterforms. It should be possible to map these consonants to existing keys on the keyboard
which are not required for other purposes.

Oromo or Galla

Of all the members of the Cushitic sub-group of the Afro-Asiatic language family, Oromo
has the most native speakers - about 11 million people, mostly in Ethiopia, and some in
Kenya. In the past it has been written in Ethiopic script, but it was not officially favoured as
a written language until 1970. However, there has long been a rich oral poetic tradition.

Consonants

Bb Cc Dd Ff Gg Hh
Jj Kk LI Mm Nn Pp
Qq Rr Ss Tt Ww
Xx Yy

Vowels

Aa Ee Ii Oo Uu

Further notes

# It may be noted that Oromo does not use V or Z. There is a total of 25 recognised consonants, five of
which are written as digraphs (e.g. ch). The glottal stop or 'qoqsa' is written with an apostrophe.

# Oromo is not a difficult language to typeset by computer. All the required characters are already in the
basic character set provided.

Somali is a Cushitic language within the Afro-Asiatic family of languages. It is spoken by
some 6 million people in Somalia, in parts of Ethiopia and Kenya, and by substantial refugee
communities abroad. The use of this simple orthography for Somali based on roman letters
was made official by the government of Siad Barre in 1972, setting aside the unique
'Osmanian' script proposed by Osman Yusuf; a fairly successful literacy campaign followed.
Some Somali-English children's picture books have been published in Britain.

Consonants

Bb Cc Dd Ff Gg Hh
Jj Kk LI Mm Nn
Qq Rr Ss Tt Ww
Xx Yy

Vowels

Aa Ee Ii Oo Uu

Further notes

# It may be noted that Somali does not use P, V or Z. An apostrophe is used to indicate the glottal stop.

# Somali is not a difficult language to typeset by computer. All the required characters are already in
the basic character set provided with all computers.

Swahili

Swahili (kiSwahili, 'coastal language') developed in Zanzibar and on the East African coast,
based on a Bantu language structure with extensive borrowings of vocabulary from Arabic
and Indian traders. Swahili was first written in 1728 in Arabic script, but later changed to
Roman letters; the first Swahili newspaper Habari ya Mwezi was published at Magila in 1895.
It has grown to become one of the principal languages of East Africa, spoken by more than
30 million people; it is the official language in Tanzania, and recognised as a secondary
language in Kenya and Uganda.

Consonants

Bb Cc

Vowels

Further notes

# Swahili does not need F, Q or X, but sometimes uses R to spell words of a European language which
use the letter, such as 'regulation'. (However, as in a similar confusion among North East Asians, many
Swahili speakers cannot distinguish between the European R and Bantu L sounds.)

# Swahili is not a difficult language to typeset by computer. All the required characters are already in the
basic character set provided.

Tswana

Tswana is a southern member of the Bantu subgroup of the Niger-Congo language family,
related to Sotho and Venda. It is spoken by about 3-3 million people in south east Africa,
especially in Botswana where it is the principal language.

Consonants

Ff Gg Hh

Mm Nn Pp

Tt Vv Ww

Vowels

Aa Ee Ee Ii Oo 06
Uu

Further notes

# At the time of writing it is not clear whether all of the standard Latin consonants shown are actually
required for Tswana, as the information was obtained from a reference source which listed only the
language's extended Latin font requirements.

• Tswana is extremely easy to typeset with existing DTP or word-processing programs for Windows or
Macintosh computers. The accented vowels E and 6 - shown in green above - are part of the
standard font encoding for these computers.

Twi

Twi - also known as Akan, Fante or Ashanti - is a member of the Kwa group of West African
Niger-Congo languages, and is spoken by 6-7 million people in Ghana and Cote d'lvoire.
The orthographic system shown below was developed by the Ghana Bureau of Languages.

Consonant range and sequence

ptkkybdggyfshhymnnngnny nny
ny ng r w w tw dw dw gu hw nw nw nu nh 1 v

Consonantal letterforms

Bb Dd Dd Ff Gg Hh Kk
LI Mm Nn Nn Nn Pp Rr
Ss Tt Vv Ww Ww Yy Yy

Vowels

Aa Aa Aa Ee Ee Ee £e

i i i i it

£e Ii II Do 05 Oo OoOo
Uu Uu

i i

Further notes

0 Like some other West African languages, Twi has a relativistic system of three tones ('tone terracing'),
but no tone markers are used in the writing system.

• The tilde mark (~) indicates nasalisation of a vowel or consonant. In common use, nasalised vowels are
usually not marked, but all possible combinations are shown above. Those letterforms marked in green
can be achieved with the standard Mac/Windows character set (h+6).

• Some of the letterforms shown could be achieved as composed characters using a typesetting system
such as T[X - especially if a round 'underdot' can be substituted for the vertical understroke glyph.
However, it is clear that a special font would be needed anyway for the two 'open' vowels.

Wolof

Wolof is a member of the West Atlantic sub-group of the Niger-Congo language family.
It is spoken by about 2-6 million people in Senegal. (The Senegalese scholar Cheikh Anta
Diop has claimed controversially that Wolof is closely related to ancient Egyptian.)

Consonants

Bb Cc Dd Ff Gg Hh
Hh Jj Kk LI Mm Nn
Pp Qq Rr Ss Tt Tt
Vv Ww Xx Yy Zz

Vowels

Aa Aa Aa Ee Ee Ee
Ii Oo Uu

Further notes

# At the time of writing it is not clear whether all of the standard Latin consonants shown are actually
required for Wolof, as the information was obtained from a reference source which listed only the
language's extended Latin font requirements.

# There are two solutions for typesetting Wolof. If a modified font is available, any word-processing
or DTP program can be used to typeset it. Alternatively, one could use a typesetting system which
can place an accent above or below any arbitrarily chosen character (the 'composed character'
approach). Any implementation of T[3< could do this with ease, but complex document layouts
such as leaflets and newsletters are hard to do in TrX

Xhosa

Xhosa is a member of the Bantu sub-group of the Niger-Congo family of languages, and is
spoken by about 6-7 million people in the north and north-east of the Republic of South
Africa. The Xhosa people merged lineages with some neighbouring Khoe peoples, and as a
consequence Xhosa has absorbed some of the Khoisan 'click'-consonant sounds.

Consonant range and sequence

b bh d f g h hi dl j k kh kr 1 m n ng ny p ph r

rh s sh t th tsh ts ty dy v w y z + clicks c q x

Consonantal letterforms

Bb Cc Dd Ff Gg Hh
Jj Kk LI Mm Nn Pp
Qq Rr Ss Tt Vv Ww
Xx Yy Zz

Vowels

Aa Ee Ii Oo Uu

Further notes

# Because it has been possible to 'recycle' three latin consonants into click-sound representations,
Xhosa can be typeset with completely standard DTP or word-processing software.

Yoruba

Yoruba, spoken by 20 million people in southern Nigeria and Benin, is one of the principal
Bantu languages in the Benue-Congo subgroup of the Niger-Congo family of languages.

Consonant range and sequence

bmftdnslrsjykgpgbwh

Consonantal letterforms

Bb Dd Ff Gg Hh Kk LI
Mm Nn Pp Rr Ss Ss Tt
Vv Ww Yy

Vowels

A a [Aa Aa] E e [Ee Ee] E e [Ee Ee]

11 ' ' ' '

Ii [Ii If] 00 [06 06] 00 [06 06]

11 4 ' ' '

Ull [Uu Uu] [Mm Mm] [Nn Nri]

Further notes

# Yoruba is a tonal language, and acute or grave accents are used over vowels to indicate tone. When
the letters M and N are used as nasalised vowels, they too may require the placement of tonal accents.
The vowel-accent combinations already provided as combined glyphs in the Windows and Macintosh
standard character sets are shown in green.

# The precise form of the 'underdot' character varies in use. Sometimes a circular dot is used, sometimes
a vertical stroke, and sometimes the vertical stroke intersects with the profile of the character above.

# All the letterforms shown could be built up as composed characters using the control codes of a type-
setting system such as T[X - especially if a round 'underdot' can be used, accessed with the \d code.
However, it is difficult to create complex layouts for leaflets and newsletters with TrX, so a special font
that supplies all of the combinations of letters and accents as prebuilt letterforms may be preferred.

Zulu

Zulu is one of the most southerly members of the Bantu sub-group of the Niger-Congo
family of languages, and is spoken by about 8-9 million people in the east of the Republic of
South Africa (KwaZulu-Natal Province). In common with some other extreme-south Bantu
languages such as Xhosa, Zulu has absorbed 'click'-consonant sounds from neighbouring
Khoisan languages.

Consonants

Bb Cc

Ff Gg Hh

Mm Nn Pp

Tt Vv Ww

Vowels

Oo Uu

Further notes

0 The Zulu language has a very rich range of consonants. Some of the 'un-clicked' Zulu consonants are
written as latin digraphs or trigraphs: e.g. sh, tsh, kh, ng.

• The three 'click' consonants in Zulu are produced by an implosive separation of the tongue from
various parts of the palette - from just behind the teeth, the middle of the hard palette, and the soft
palette at the back of the mouth. The letters C, Q and X are used to represent the simple form of these
clicks, which can also be aspirated (ch, qh, xh), nasalised (nc, nq, nx) or voiced (pc, gq, gx).

# Because it has been possible to 'recycle' three latin consonants into click-sound representations,
Zulu can be typeset with completely standard DTP or word-processing software.

Internet Archive Audio

Featured

Top

Images

Featured

Top

Software

Featured

Top

Texts

Featured

Top

Video

Featured

Top

Mobile Apps

Browser Extensions

Archive-It Subscription

Save Page Now

Full text of "Typesetting African Languages"

See other formats